Digital video for the documentation of colonoscopy

Effectiveness of asynchronous tele-endoscopy

BACKGROUND

Asynchronous tele-endoscopy can improve access and quality of patient care. This is the first published evaluation of the diagnostic accuracy of highly compressed digital video in GI endoscopy.

OBJECTIVE

To determine whether asynchronous tele-endoscopy using highly compressed video can accurately document and diagnose lesions in the upper GI tract.

DESIGN

Local endoscopists performed 50 elective upper GI endoscopies. A high-quality DV compressed video (25 megabits per second [Mbps], 720 x 480 pixels) and highly compressed MPEG-1 video (2.0 Mbps, 352 x 240 pixels) were simultaneously captured. Five endoscopists asynchronously reviewed 20 compressed digital videos (100 case reviews) for endoscopic diagnoses. In addition, demonstration technique and image quality were rated on a Likert scale. Concordance between local and asynchronous endoscopists for major and minor endoscopic findings was evaluated. An independent panel classified discrepancies as caused by image quality, endoscopic technique, or interobserver variability through comparison of the 2 forms of digital video.

RESULTS

Although asynchronous endoscopists rated the image quality of highly compressed video as diagnostic in 85% of cases, only 18% of studies yielded the same clinical diagnoses. There was high discordance for both major (kappa = 0.38, 95% CI, 0.19-0.57) and minor findings (kappa = -0.29, 95% CI, -0.43 to -0.15). Interobserver reporting was responsible for 90% of variability in contrast to only 4.9% for poor image quality.

CONCLUSIONS

The findings suggest that the diagnostic accuracy of low-bandwidth, low-resolution, highly compressed video is well tolerated and comparable to the current standard. Interobserver reporting variability accounted for most of the poor correlation. Improved synoptic documentation is required for effective communication among endoscopists.

[Sampling studies in ambulatory quality assurance -- using the example of colonoscopy]

BACKGROUND

Sampling inspections are an approved instrument for assuring and promoting the quality of healthcare. Individual documentations of medical services are requested from physicians and randomly selected for quality rating by experienced peer reviewers. For example, sampling inspections are stipulated by law for certain ambulatory services (e.g., colonoscopies) delivered by SHI-authorised physicians in order to maintain professional performance standards of colonoscopists.

OBJECTIVES

On behalf of the regional Association of Statutory Health Insurance Physicians (ASHIP) experienced colonoscopists regularly rate selected visual documentations (videotapes, photographs) of colonoscopies performed by SHI-authorised physicians in the ambulatory care sector. For anatomical reasons, however, a certain proportion of colonoscopies of inadequate quality will generally be tolerated. Whenever this value is exceeded, ASHIP may impose sanctions against the physician. The question is how sampling inspections have to be performed and dimensioned in order to ensure that the tests be sufficiently meaningful in terms of sensitivity and specificity.

METHODS

Relevant sampling test parameters such as the false-positive rate (physicians are wrongly accused of inadequate quality) and the false-negative rate (existing deficiencies are not identified) are calculated. The calculations are performed analytically or, in the case of complex sampling test scenarios, numerically by means of computer simulations.

RESULTS

The calculations show that single-stage sampling tests usually do not result in acceptable values for the false-positive and the false-negative rates. For example, a sampling test which requires the documentation of 20 colonoscopies will -- assuming some reasonable tolerance of inadequately performed colonoscopies -- result in a false-positive rate of 6% and a false-negative rate of 47%. The false-positive-rate, which is particularly relevant from a legal point of view, can be reduced by providing a two-stage sampling test. A significant reduction of the false-negative-rate may be achieved by (multiple) repetition of the single-stage sampling test and consideration of cumulative probabilities.

CONCLUSIONS

In principle, sampling inspections permit statements in terms of probabilities only. In sampling inspections of healthcare quality false-negative rates are usually considered, i.e., the probability that the test is unable to identify existing quality deficiencies. However, false-positive rates also need to be considered in cases where sanctions may be imposed against the physician on the basis of a positive sampling test. Numerical calculations of false-positive and false-negative rates for simple and complex sampling test scenarios should be performed in order to choose the optimum procedure and dimension of a sampling test.

What are the sensitivity and specificity of endoscopic photographs in determining completion of colonoscopy? Results from an online questionnaire

OBJECTIVES

There is no consensus as to the optimal method for documenting colonoscopy completion. For endoscopic photographs to be useful for documentation, reviewers must be able to distinguish between photographs taken from incomplete and complete colonoscopy. The primary aim of this study was to calculate the diagnostic specificity and sensitivity of a pair of photographs in confirming complete colonoscopy. The secondary aim was to assess current methods of documenting colonoscopy completion.

METHODS

Eighty pairs of photographs were taken from completed colonoscopies. Two photographs, including at least one view of the ileo-caecal valve, were taken of caecal landmarks and/or the terminal ileum. Colonoscopy completion was independently validated using video clips. Twenty pairs of photographs were also taken from another colonic site that could potentially be misinterpreted as the caecum, for example, hepatic flexure. Using an online questionnaire 34 experienced endoscopists were asked about their own methods of colonoscopy documentation. Thirty-two of these reviewers then assessed the 100 photographic pairs, blinded to their origin, and were asked 'Taking both photographs into account, are you convinced that complete colonoscopy has been performed?'

RESULTS

Fifty-six percent of endoscopists did not routinely use any objective method of documentation although 38% frequently used some photographs for documentation. Using a pair of endoscopic photographs to document colonoscopy completion has a sensitivity of 51.4% (confidence intervals: 49.5-53.3%) and a specificity of 89.2% (confidence intervals: 86.8-91.6%).

CONCLUSION

Both the sensitivity and the specificity of a pair of endoscopic photographs are too low to be used for reliably documenting colonoscopy completion.

The functionality and reliability of an Internet interface for assessments of endoscopic still images and video clips: distributed research in gastroenterology

BACKGROUND

Validation studies of medical images are required before the general use of new imaging techniques, and they necessitate a substantial number of observers. Multicenter studies are desirable but the logistic challenges are considerable.

OBJECTIVE

The aim of the present study was to develop and test the functionality and reliability of an Internet interface for distributed evaluation of endoscopic still images and video clips.

DESIGN

An Internet site for assessment of endoscopic still images and video clips was developed. To avoid limitations of the bandwidth, a solution was developed to launch and score high-quality video clips on digital video disks (DVDs) from the Internet interface.

SETTING

Fifty-five observers were enrolled in the randomized, prospective multicenter study of still images. The feasibility study of video clips included 7 observers.

PATIENTS

Four patients presenting with various degree of ulcerative colitis, ranging from normal to severe, were included.

MAIN OUTCOME MEASUREMENTS

We tested the reliability of the interface by dual-image assessment, as well as the feasibility of Internet assessment of endoscopic images. We physically distributed high-quality DVD video footage.

RESULTS

We recorded 2084 assessments of endoscopic still images and 35 assessments of video clips. The reliability of the Internet interface was confirmed by adequate repeatability and intraobserver agreement of the assessments. Video clips running from a DVD were also successfully shown on the Internet interface. Thus, high-quality video may be assessed independently of the Internet bandwidth.

CONCLUSIONS

The present Internet-based tool is functional, efficient, and reliable for high-volume assessment of endoscopic images and video clips.